Bring your own data

Day 4 - Introduction to Data Analysis with R

Selina Baldauf

Freie Universität Berlin - Theoretical Ecology

October 17, 2023

Organization

Schedule of today

  • Now - 14 (or 14.30 if you are enthusiastic still): Work on the data set(s)

    • Take break(s) as best fits your needs

  • 14 (14.30) - 15: Short feedback round

    • What did you find out about your data set? Plots, summaries, …
    • Which methods did you use?
    • Did you learn something new?
    • Was there something you struggled with?
    • …

  • 15-16: Feedback, conclusion

New methods: Correlation plots

New methods: Correlation plots

New methods: PCAS

New methods: Correlation plots and PCAS

Data set 1: What makes a good wine?

Physicochemical properties of wine and quality judgements

'data.frame':   1599 obs. of  12 variables:
 $ fixed.acidity       : num  12.7 9.8 6.5 8.6 7.5 7.6 10.1 6.4 6.1 6.7 ...
 $ volatile.acidity    : num  0.6 0.66 0.88 0.52 0.58 0.5 0.935 0.4 0.58 0.46 ...
 $ citric.acid         : num  0.49 0.39 0.03 0.38 0.14 0.29 0.22 NA 0.23 0.24 ...
 $ residual.sugar      : num  2.8 3.2 NA 1.5 2.2 2.3 3.4 1.6 2.5 1.7 ...
 $ chlorides           : num  0.075 0.083 0.079 0.096 0.077 NA 0.105 0.066 0.044 0.077 ...
 $ free.sulfur.dioxide : num  5 21 23 5 27 5 11 5 16 18 ...
 $ total.sulfur.dioxide: num  NA 59 47 18 60 NA 86 12 70 34 ...
 $ density             : num  0.999 0.999 0.996 NA 0.996 ...
 $ pH                  : num  3.14 3.37 NA 3.2 3.28 3.32 3.43 3.34 3.46 3.39 ...
 $ sulphates           : num  0.57 0.71 0.5 0.52 0.59 NA 0.64 NA NA 0.6 ...
 $ alcohol             : num  11.4 11.5 11.2 9.4 9.8 11.5 11.3 9.2 12.5 10.6 ...
 $ quality             : int  5 7 4 5 5 6 4 5 6 6 ...

Data set 1: What makes a good wine?

Ideas - know methods

  • Plot of wine quality against chemical properties
  • Plot of distribution of chemical properties
  • Summary tables using dplyr

Ideas - new methods

  • Correlation plots: How are the different wine properties correlated with each other?
  • PCA: How are the wine properties related to each other?

Frederik Vandaele - originally posted to Flickr as Château Pétrus, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=5145286

Data set 1: What makes a good wine?

Hints

  • Transform the quality column to a factor before plotting: use dplyr::mutate and as.factor() to tranform the column
  • Try the janitor::clean_names() function

Frederik Vandaele - originally posted to Flickr as Château Pétrus, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=5145286

Data set 2: Paralympic games from 1980-2016

Most important variables:

variable class description
gender character Binary gender
event character Event name
medal character Medal type
athlete character Athlete name (LAST NAME first name
abb character Country abbreviation
country character Country name
type character Type of sport
year double year of games

Data set 2: Paralympic games from 1980-2016

Get the data:

athletes <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2021/2021-08-03/athletes.csv')


Ideas - know methods

  • Create summaries of medal counts for different groups with dplyr
  • Did the ratio of men/women winning medals change over time?
  • Which countries were the most successful ones? Does this differ between sports type?
    • Which types of sports accumulated the most medals?
  • Make plots such as:
    • Age distribution of athletes winning gold, silver and bronze
    • Compare the total number of medals over the years between winter and summer Olympics

Data set 2: Paralympic games from 1980-2016

Hints

  • To reduce complexity of the data, first filter only the athletes that won a medal (!is.na(medal))

Data set 3: Crab data set

Atlantic marsh fiddler crab (Minuca pugnax)

  • Crab from Florida is expanding northward due to ocean warming
  • Data on 13 marshes across a range of latitude in the USA
  • Recording of the size of of the crab
  • Rather small and good to handle

Image by LTER under CC BY-SA 4.0

Data set 3: Crab data set

# A tibble: 6 × 9
  date       latitude site   size air_temp air_temp_sd water_temp water_temp_sd
  <date>        <dbl> <chr> <dbl>    <dbl>       <dbl>      <dbl>         <dbl>
1 2016-07-24       30 GTM    12.4     21.8        6.39       24.5          6.12
2 2016-07-24       30 GTM    14.2     21.8        6.39       24.5          6.12
3 2016-07-24       30 GTM    14.5     21.8        6.39       24.5          6.12
4 2016-07-24       30 GTM    12.9     21.8        6.39       24.5          6.12
5 2016-07-24       30 GTM    12.4     21.8        6.39       24.5          6.12
6 2016-07-24       30 GTM    13.0     21.8        6.39       24.5          6.12
# ℹ 1 more variable: name <chr>

Data set 3: Crab data set

Ideas - known methods

  • Explore Bergmann’s rule (organisms are large in higher latitudes)
  • t-tests to compare size between locations
  • Plot relationship between latitude and size
  • Plot distributions of variables

Image by LTER under CC BY-SA 4.0

Data set 4: Ice cover and temperature

Temperature and ice duration on lakes since 19th century

  • 2 data sets with measurements of
    • ice start, end and duration on 2 lakes in Wisconsin
    • daily air temperature since 1870
  • Explore the effect of climate change on ice cover

Image by LTER under CC BY-SA 4.0

Data set 4: Ice cover and temperature

Ice data:

# A tibble: 6 × 5
  lakeid       ice_on     ice_off    ice_duration  year
  <fct>        <date>     <date>            <dbl> <dbl>
1 Lake Mendota NA         1853-04-05           NA  1852
2 Lake Mendota 1853-12-27 NA                   NA  1853
3 Lake Mendota 1855-12-18 1856-04-14          118  1855
4 Lake Mendota 1856-12-06 1857-05-06          151  1856
5 Lake Mendota 1857-11-25 1858-03-26          121  1857
6 Lake Mendota 1858-12-08 1859-03-14           96  1858

Temperature data:

# A tibble: 6 × 3
  sampledate  year ave_air_temp_adjusted
  <date>     <dbl>                 <dbl>
1 1870-06-05  1870                  20  
2 1870-06-06  1870                  18.3
3 1870-06-07  1870                  17.5
4 1870-06-09  1870                  13.3
5 1870-06-10  1870                  13.9
6 1870-06-11  1870                  15

Data set 4: Ice cover and temperature

Ideas - known methods

  • How did ice cover duration change over the years?
  • How did air temperature change over the years?
    • Summarize mean annual temperature or mean temperature in winter
  • How do ice duration on the lakes correlate with temperature (e.g. with mean winter temperature)

Image by LTER under CC BY-SA 4.0

Hints

  • For some plots it might make sense to summarize annual means first
  • Use left_join to combined the tables with annual mean temperature and ice duration

Data set 5: RNAseq data

  • Data from FU et al. 2015, Nature Cell Biology

  • Data found via Tutorial on heat maps using this data

  • 3 csv files:

    • heatmap_genes.csv: A list of the names of interesting genes to look at (Genes used in Figure 6b in paper)
    • DE_results.csv: Gene expression in luminal cells in pregnant versus lactating mice
      • logFC, AveExpr, t, p-value
      • Also contains non-significantly expressed genes
    • normalized_counts: Normalized counts for genes for the different samples

Data set 5: RNAseq data

Ideas:

  • Create a heatmap of the top 20 most significant genes (see plot in the tutorial)
  • Create a heatmap of the interesting genes (see Fig. 6 in the paper)
  • Create a volcano plot of the data similar to the one here

Data set 5: RNAseq data

Data set 5: RNAseq data

Some tips:

Data cleaning:

  • Read in the data and then use the janitor::clean_names function to make the column headers nicer
  • Join DE_results and normalized_counts by their shared columns
  • Use select to remove columns you don’t need for analysis to get a better overview
  • Filter only significant genes (tutorial) defines them as p_value < 0.01 & abs(logFC) > 0.58)

Data analysis:

  • Heatmap with ggplot or with pheatmap::pheatmap()
    • pheatmap takes a matrix as input (use as_matrix on tibble to transform)
  • scale the counts -> have a look at the scale function
    • pheatmap can scale but with ggplot you have to scale before plotting

Some general tips

  • First make a plan:
    • What do you want to achieve and what are the steps?
    • Try to think in technical terms
  • Start with something small, e.g. reading in the data and bringing it into the right format.
  • Google
  • If you get stuck, ask in the chat or stop by in General
  • Have a look at the package cheat sheets

Now you

Working with real research data

Meet in your group (if you want)
Work on your data set
Take breaks as you need and be back at 2 p.m.
Keep an eye on your group and the general chat

Sharing

In 1-2 mins:

  • What was the highlight of your analysis?

    • Your favorite plot
    • Some cool code
    • A problem that you finally solved
    • Something new you learned

  • What was difficult?

  • If you want: Share a screenshot in the chat or share your screen

Feedback

Please take 10 mins to complete the feedback survey for the Graduate center (don’t use Internet Explorer)

https://votingo.cedis.fu-berlin.de/PCNLP3

Feedback

  • Any other feedback or comments from your side?

Conclusion

We learned a lot of stuff!

Conclusion

How to continue from here?

Artwork by Allison Horst

The End

Thanks a lot for participating!

Artwork by Allison Horst

Selina Baldauf // Bring your own data